viral-marketing

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • PROMPT_INJECTION (SAFE): Instructions focus on domain-specific grounding and persona definitions. No attempts to subvert safety filters or ignore system rules were detected.
  • DATA_EXFILTRATION (SAFE): The skill does not access sensitive local paths (like .ssh or .aws) and makes no network requests.
  • REMOTE_CODE_EXECUTION (SAFE): No code execution commands, script downloads, or shell invocations are present.
  • INDIRECT_PROMPT_INJECTION (SAFE): The skill has a validation surface for processing user-provided marketing text. Evidence Chain: 1. Ingestion: Marketing copy in .md and .html files. 2. Boundaries: None. 3. Capabilities: None (no exec, network, or disk write). 4. Sanitization: None. The surface is not exploitable due to the absence of capabilities.
  • DYNAMIC_EXECUTION (SAFE): No runtime code generation, compilation, or unsafe deserialization was found.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:58 PM